|Publication number||US5771102 A|
|Application number||US 08/553,467|
|Publication date||Jun 23, 1998|
|Filing date||Mar 8, 1995|
|Priority date||Mar 11, 1994|
|Also published as||DE4408163A1, EP0698316A1, EP0698316B1, WO1995024770A1|
|Publication number||08553467, 553467, PCT/1995/320, PCT/DE/1995/000320, PCT/DE/1995/00320, PCT/DE/95/000320, PCT/DE/95/00320, PCT/DE1995/000320, PCT/DE1995/00320, PCT/DE1995000320, PCT/DE199500320, PCT/DE95/000320, PCT/DE95/00320, PCT/DE95000320, PCT/DE9500320, US 5771102 A, US 5771102A, US-A-5771102, US5771102 A, US5771102A|
|Inventors||Wilhelm Vogt, Dirk Lappe, Rodolfo Mann Pelz, Oliver Vogt, Gunnar Nitsche, Frieder Mundt|
|Original Assignee||Robert Bosch Gmbh|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (10), Non-Patent Citations (2), Referenced by (21), Classifications (28), Legal Events (7)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The invention is directed to a process for the transmission of images in which data are generated by at least one data source, are inserted in a data frame at a channel coder and are subjected to a channel coding. A process for the transmission of images is described in Biere, "Ubertragung von Bewegt- und Standbildern in stark gestorten schmalbandigen Kanalen beim Funkbildtelefon Transmission of moving images and still images in narrow-band fax telephone channels with high interference!", Bosch Technische Berichte, No. 54, 1991, in which the image data of an image undergo a source coding and the source-coded image data are then channel-coded and transmitted via a data channel. The source coding is effected in the form of a predictive error coding and a block-oriented movement estimation. A rate-compatible punctured convolutional code (RCPC) is used for channel coding, the occurring data streams being punctured and subsequently interleaved to form a common data stream.
The process according to the invention enables various kinds of data to be channel-coded and transmitted at various code rates. Information about the code rate and the class type is advantageously stored in an information word. This enables an efficient dividing up of the available data rates.
In a particularly advantageous manner, information about the type of data generation is transmitted itself as data of a class. In particular, image data, audio data, facsimile data, and data in the form of data files are transmitted. The channel error protection is improved by using an interleaving method. By forming a data frame with multiple whole-number storage locations to be found in a diagonal of a memory matrix, it is possible to find the synchronizing word quickly by means of the data interleaving method and to carry out an accurate synchronization of the deinterleaver and data frame.
Instead of code information, additional data is advantageously entered in the information word. Accordingly, data can be exchanged from the channel coder to the channel decoder without having to change the data protocol. A Viterbi decoder which operates according to the Maximum Sequence Estimation process is particularly suitable for channel decoding.
The process according to the invention for transmitting image data has the advantage over the prior art that the source-coded data are divided into classes, these classes indicating, e.g., the sensitivity of the source-coded image data to transmission errors and are coded with different code rates according to the extent of the error protection. These classes are provided with varying degrees of error protection depending on sensitivity, the sensitive image data receiving greater error protection than the less sensitive image data. In this way, the available data rate is used more efficiently. Accordingly, it is possible to improve image quality at the same data rate. The division of the image data into classes and the length of the classes and error protection used are transmitted by way of a first information word so that these data which are required for decoding are transmitted also and made available to the decoder. Data relating to the source coding are preferably included in the transmission in the form of code words. A second information word is used to identify the length of the code words, the class of the code words, and the error protection of the code words. In this way, additional information is made available for the source decoding. The quality of the source-decoded image data is accordingly improved.
Further embodiments of the preferred process provide additional improvements over the prior art. In a particularly advantageous manner, a synchronizing word composed of two Williard code words or two Barker code words is used for synchronization. The synchronizing word is added to the image data in a fixed pattern as a pseudo-noise (PN) synchronizing word. This enables a simple and accurate synchronization with sufficient protection against transmission errors.
The information word or the first and second information word for each class advantageously includes class number, a code number, and a length number. The class number provides the class of the associated code words or the class of the associated image data. The code number contains the code rate used for the channel coding and the length number provides the quantity of bits associated with the code words or class.
Since the first and second information words are terminated by a predetermined number of zero bits, the Viterbi decoder is in a known state. The information word or the first and second information word can be evaluated subsequently without having to evaluate additional data bits by inserting data bits which do not change the state of the Viterbi decoder. Subsequently, the following code words or the following image data are decoded using the data contained in the information word or in the first and second information words.
The code words preferable comprise a first and second code word, the first code word having a length of three bits where the first bit of the code word indicates whether the image data are formed quickly or slowly when the image is reproduced. The second bit of the first code word indicates which of two possible source coding methods was used. The third bit of the first code word indicates whether an information word or a first and second information word is used. The second code word preferably has a length of five bits and contains the step value of the quantization used in the source coding.
It is advantageous to use a third and a fourth code word, where the third code word has a length of one bit and indicates whether or not additional information relating to the type of source coding is stored in the following fourth code word which has a length of eight bits. In this way, it is possible to transmit additional information concerning the source coding by means of the third and fourth code words and to use it for image source decoding.
The source-coded image data concerning contour, movement vector, block position, DCT coefficient, quantizing step, and structure index are preferably divided into different classes. This ensures an efficient transmission of image data and enables an improved image quality at the same data rate.
The flexibility of the described process is increased in that the code number has unoccupied codes so that codes will be available for future classes. In this way, it is possible to use the data protocol to process new classes in future image processing methods having new classes without changing the data protocol. The described process is accordingly well suited for further developments.
The process is improved in that the length index or length pointer is used not only to indicate the amount of image data bits contained in a class, but also as an indicator for the decoder to indicate whether all of the image data bits of a class are contained in this class or whether another class with an additional first information word will follow. This increases the processing speed of the decoder.
An advantageous construction of the interleave consists in the use of a convolutional interleave having a memory matrix with a fixed width and an infinite length. The data bits are read in diagonally from top left to bottom right and the successive diagonals are offset horizontally relative to one another by one storage location. For the purpose of transmitting image data, the image data bits are read out of the memory matrix from successive lines in the sequence by which the diagonals were offset relative to one another when read in. This enables a simple and fast interleaving process which offers good protection against burst errors in the data channel.
It is particularly advantageous to fix the amount of data bits to be transmitted for an image at a multiple of the bits which can be read into a diagonal in the matrix, the missing image data bits being replaced by empty bits. In this way, it is possible for the image data bits to be read into the matrix memory of the deinterleaver in such a way that the data bits of the synchronizing word occupy a determined column and are accordingly in a predetermined grid in the second memory matrix of the deinterleaver. The synchronizing word is found with the help of a correlating process. It is accordingly possible to synchronize all system components of the receiver with one synchronizing word.
In a preferred imbodiment of the method of the invention a first information word is dec according to a fixed structure, with a fixed bit length and a fixed code rate. The following image data bits are subsequently channel decoded corresponding to the information stored in the first information word. This enables a fast and efficient channel decoding.
A second information word is preferably channel-decoded with a fixed structure, a fixed bit length and a fixed channel coding and information concerning the type of class of the following code words, the type of error protection, i.e., the code rate, and the amount of data bits belonging to the code words is determined from the second information word in order subsequently to channel-decode the code words following the second information word using the data stored in the second information word and to use the information of the code word for subsequent image source decoding of the image data bits.
A convolutional deinterleaver having a memory matrix is advantageously used to implement the eeinterleaving process. The received data are read in diagonally from the bottom left-hand side to the top right-hand side, successive diagonals are offset relative to one another by one storage location in the vertical direction, and the data are read out line by line subsequently. The lines are offset relative to one another by one storage location and in the same direction as the diagonals in which the data are read in. In this way, it is possible to carry out a fast and simple deinterleaving process which simultaneously provides the data to be transmitted with adequate protection against burst errors in the data channel.
In a particularly advantageous manner, a Viterbi decoder, i.e., a maximum likelihood symbol estimation decoder, based on the Viterbi algorithm is used for channel decoding. The Viterbi decoder has a fixed memory depth. After the information word or the first or second information word is decoded, the Viterbi decoder is in a known state and the information word or the first or second information word can be evaluated by a quantity of entered zero bits or one bits. The data following the information word or following the first or second information word are then channel-decoded by the Viterbi decoder with the information contained in the information word or in the first or second information word. This enables an efficient transmission process in which data transmission and channel decoding are adapted to one another with respect to time so that the data of the first and second information word are determined before the image data are channel decoded.
The objects, features and advantages of the invention will now be illustrated in more detail with the aid of the following description of the preferred embodiments, with reference to the accompanying figures in which:
FIG. 1 is a block diagram of apparatus for processing, transmitting and receiving image data used to perform the process according to the invention;
FIG. 2 is a diagram of a data frame of an image;
FIG. 3 is a diagram of a synchronizing word;
FIG. 4 is a diagram of an information word;
FIG. 5 is a diagram showing code words used in the process according to the invention;
FIG. 6 is a table showing the classification of source-coded image data in relation to code rate for channel coding;
FIG. 7 is a table showing a code pointer for the first or second information word;
FIG. 8 shows a convolutional interleave memory matrix and convolutional deinterleaver memory matrix;
FIG. 9 is a flow chart for a process of transmitting images according to the invention;
FIG. 10 is a diagram of an image divided into different regions designated by numbers according to the different classes of error protection performed in the method according to the invention; and
FIG. 11 is a flow chart for a process of receiving and decoding channel-coded and source-coded image data according to the invention.
FIG. 1 shows a digital signal processor 1 which is connected with a multiplexer 50 via a data bus 26. The multiplexer 50 is connected with an image data source 51 (image), an audio source 52 (audio), and a telephone facsimile 53 (fax). The data sources feed digital data to the first digital signal processor 1. The digital signal processor 1 is connected with a transmitter unit 2 via a transmission line 27.
FIG. 1 also shows a receiver unit 4 which is connected via a reception line 28 with a second digital signal processor 3. The second digital signal processor 3 is connected with an image sink 54, an audio sink 55 and a facsimile sink 56 via a second data bus 29. The second digital processor 3 sends the received data to the appropriate data sinks.
FIG. 2 shows a data frame 5 having a fixed quantity of data bits. The data frame 5 has a synchronizing word 6, a second information word 7, code words 8, a first information word 9, image data bits 10, and termination bits 30. A data frame contains, in each instance, the data required for transmission of an image. In general, the data frame 5 includes a plurality of first and second information words 7, 9, a plurality of code words 8 and a plurality of classes of image data bits 10. For the sake of simplicity, not all data are shown in FIG. 2. For the transmission of data according to the claim 1, the data frame is constructed in an appropriate manner, the information word being represented by the first or second information word. No distinction is made between the first and second information word in claim 1.
A synchronizing word 6 which has a length of forty bits and represents a pseudo-noise combination of two Williard words is shown schematically in FIG. 3. A pseudo-noise combination of two Barker words could also be selected. A combination of Williard words or Barker words is used because the known Williard words and Barker words only have a length of up to 33 bits, which amount is not always sufficient to enable an accurate synchronization with adequate error protection. The synchronizing word 6 is inserted in the data frame 5 according to a fixed pattern and a bit of a synchronizing word is always in the first position of the data frame.
FIG. 4 shows a first or second information word 9, 7 since the first and second information words 9, 7 are formed identically. The first or second information word 9, 7 has a class number 12 with a length of four bits, a code pointer 13 with a length of five bits, a length number 14 with a length of fourteen bits, and six zero bits 15. The class number 12 indicates the class to which the following image data 10 belong. The code pointer 13 contains the information concerning the error protection with which the image data 10 are channel coded, i.e., the code rate of the channel-coded image data 10. The length number 14 indicates the quantity of image data bits 10 or the quantity of the data bits belonging to the code words 8.
The quantity of zero bits 15 which terminate each information word is adapted to the memory depth of the channel coder. Since the channel coder used has a memory depth of six bits, six zero bits 15 are used.
FIG. 5 shows four code words 8, of which a first code word 16 has a length of three bits, a second code word 17 has a length of five bits, a third code word 18 has a length of one bit, and a fourth code word 19 has a length of eight bits. The first bit of the first code word 16 indicates whether the image data are produced quickly or slowly when the image is reproduced. The second bit of the first code word 16 indicates which of two possible source coding methods was applied. The third bit of the first code word 16 indicates whether or not information words are used.
The second code word 17 indicates the step value of the quantization used in the image source coding. The third code word 18 indicates whether additional information concerning the image source coding which is only to be used in future developments is stored in the following fourth code word 19.
The code words 8 are used to transmit information concerning the image source coding with the image data and to use this information in the image source coding.
FIG. 6 shows the classification of the source-coded image data with the corresponding code rates used for channel coding. The first class is allocated to the movement vectors MV. The second class is allocated to the DCT coefficients (DCT). The third class is allocated to the quantization (QU). The fifth class is occupied by the position data (P) of the macroblocks.
FIG. 7 shows a table for the code pointer 13 of the first or second information word 9, 7. Under H, information concerning the total image is represented by the code 0000, the movement vectors MV are represented by the code 0001, the macroblocks B are represented by code 0010, empty bits L are represented by code 0011, and the H0.261 process is represented by code 0100. Also shown in FIG. 7 are unoccupied codes which can be used for future classes.
FIG. 8 shows a data stream 22 formed by a sequence of synchronizing bits SI, user bits NI, and termination bits 30. The data stream represents part of a data frame. The user bits are represented, e.g., by the first or second information word 9, 7, the code words 8 or the image data bits 10. Further, a first memory matrix 23 of an interleave is shown. The data stream 22 is read into the first memory matrix 23 of the interleave in the form of diagonals from top left to bottom right. The data bit stream 22 shows the data bits necessary for transmitting an image. The quantity of termination bits A is so selected that entire diagonals are filled up with the data of an image when the data bits are read into the first memory matrix 23 of the interleave. The data bits of the first memory matrix 23 are read out again line by line from top to bottom and are transmitted to a receiver unit 3 via a data channel by a transmitter unit 2. The receiver unit 3 transmits the obtained data to the second signal processor 3 which reads the received data stream into a second memory matrix 24 of a deinterleaver corresponding to FIG. 8. FIG. 8 shows a second memory matrix 24 of a deinterleaver into which the data of the first memory matrix 23 of the interleave are written in diagonals from bottom left to top right. The data of the second memory matrix 24 are read out line by line from top to bottom so that the original data signal 22 occurs again and the data bits are again made available in the correct time sequence.
The program flow of the process for the transmission of images is shown schematically in FIG. 9. The image is divided into macroblocks at program point 31. The image is in CIF or QCIF format, for instance. The CIF format has 352 pixels in a line and 288 lines. The QCIF format has 176 pixels in a line and 144 lines. An image is divided into 99 macroblocks, each macroblock having four luminance blocks and two chrominance blocks. Each of the four luminance blocks and two chrominance blocks has 64 pixels (eight times eight).
The image data of the first macroblock which is arranged at the upper left side of the image are subjected to a source coding at program point 32. In so doing, the macroblocks are subjected to a block-by-block movement estimation, the movement vectors MV being determined in relation to a precursor image. The movement vectors MV are transmitted to the chrominance blocks in an analogous manner. A discrete cosine transform is carried out at program point 33 either intra-mode, i.e., a discrete cosine transform of the macroblock is effected in the original block, or inter-mode, where a discrete cosine transform is effected in the difference between the original image and a predicted image. A discrete cosine transform is carried out in a two dimensional manner on the luminance blocks and chrominance blocks with a value of eight times eight pixels and DCT spectral coefficients are obtained in this way. The DCT spectral coefficients are quantized and subjected to a zigzag run length coding. The discrete cosine transform is carried out as an intra-mode discrete cosine transform as was described above. Another process consists in carrying out an inter-mode discrete cosine transform.
Suitable source coding methods are known and are described, for instance, in PTT Research Neher Laboratories The Netherlands, "Description Reference Model 8" (RM8), Paris, 1989.
The contour, movement vectors MV, position of the luminance and chrominance blocks, DCT spectral coefficients, and the step value of the quantization of the spectral coefficients are accordingly available as source-coded image data which must be transmitted. Additional source coded image data comprise information on the quantization step of the differential quantization (DQUANT), for example. For this purpose, a word with a length of two bits indicates the change in quantization. Other information specifies the luminance blocks which are contained in a macroblock and which contain coded transform coefficients (CBPY). In addition, information specifying the chrominance blocks of a macroblock which contain coded transform coefficients, the prediction mode used for the image or the macroblock, and the quantization used for the transform coefficients (CBPCM) is transmitted. In addition, a code word (COD) indicates whether or not coding is carried out.
These source-coded image data shown in FIG. 6 are divided into classes. The classes are determined according to the sensitivity of the source-coded image data in relation to transmission errors. Code rates are allocated to the classes but not directly linked with the classes. Identical classes can also have different code rates. Image data which must be provided with special protection against transmission errors are assigned a low code rate. FIG. 6 shows an allocation of source-coded image data to corresponding classes and code rates. The classes are intended as information on the data sinks or image source coders.
After the source-coded image data are divided into a plurality of classes, the first information word 9 for the classes to be transmitted is established at program point 34. The class number 12 in FIG. 4 indicates class 1. If 24 data bits belong to class 1, the length pointer 14 indicates the number 24 in the form of a binary number as is shown in FIG. 4. The quantity of zero bits 15 is fixed at six zero bits in all first and second information words. If the quantity of data bits needed for transmitting the source-coded image data of a class exceeds the number 214 -1, the length pointer receives the value 214 -1 and another first information word is formed which is followed by the missing source-coded image data of the preceding class. The decoder accordingly knows that a length pointer having the value 214 -1 is followed by an addition to the preceding class. An analogous procedure is followed for the following classes of the source-coded macroblock and a first information word 9 is prepared in each instance.
An inquiry is then made at program point 35 as to whether or not all macroblocks of the image to be processed have been source-coded. If this is not the case, the program skips back to program point 32 and the following macroblock is source-coded. The macroblocks are successively source-coded line by line from top left to bottom right. However, if all macroblocks of the current image are surce-coded, the program jumps to program point 36.
At program point 36, the code words 8 containing the information on the image source coding are prepared. If an image is produced quickly, the first bit of the first code word 16 is a zero. A first source coding process was used during the actual image source coding so that the second bit of the first code word 16 is likewise zero. The third bit of the first code word 16 has a one, since information words are used. The second code word 17 indicates the step value of quantization so that the second code word 17 is to be formed as a corresponding binary number. The third code word 18 is occupied by a zero, since no additional information on the source coding process is stored in the fourth code word 19. The fourth code word is subsequently occupied by eight zeros. FIG. 5 shows the code words 8 which have just been described.
A second information word 7 which is structured in a manner corresponding to the first information word 9 and is used to identify code words 8 is then formed at program point 37. The second information word 7 gives the class, the code rate, and the quantity of data bits of the code words 8 and contains six terminating zero bits 15. As in the case of the first information word, this is indicated by a class number 12, a code pointer 13, and a length number 14. The digital signal processor 1 then prepares a synchronizing word with a length of forty bits at program point 38 as is shown in FIG. 3 and inserts the synchronizing word 6 in the data frame according to the channel coding corresponding to a predetermined pattern. Combinations of known Williard words and known Barker words are used for this purpose.
The digital signal processor 1 produces a data frame from the data, this data frame including a synchronizing word 6, a second information word 7, code words 8, a first information word 9 for the first class 1, and corresponding first information words and source-coded image data for the additional classes of the source-coded image, and a fixed quantity of termination bits 30. The quantity of data bits belonging to an image is fixed at a multiple of a predetermined quantity by means of the termination bits 30. The predetermined quantity is determined by the amount of image data which can be written into the memory matrix of the interleave in a diagonal so that the image data of an image fill up an entire quantity of diagonals in the memory matrix of the interleave.
At program point 39, the digital signal processor 1 carries out a channel coding of the image data needed to transmit an image. The second and first information words are channel-coded with the highest degree of error protection. The code words 8 are channel-coded with the error protection stored in the second information word in the code pointer. The source-coded image data are channel-coded with the error protection which corresponds to their class and is stored in the code pointer 13 in the first information words 9. A convolutional coding is preferably used for channel coding.
Convolutional coding is known and described in detail in Clark, "Error Correction Coding for Digital Communications", Plenum Press, New York, 1981, ISBN 0-306-40615-2, 227 ff. Additional coding procedures are described on pages 267 ff.
The channel-coded data of an image are then sent to the transmitter unit 2 at program point 40 and the transmitter unit 2 transmits the channel-coded image data by frequency modulation or GSFKI modulation. After program point 40, the program jumps back to program point 31 and the next image is processed. If delays occur in the channel coding so that there is no image data available for transmission when data must be transmitted, a zero bit is added and transmitted by the first digital signal processor 2.
If it is desired to install protection against burst errors in the data channel, an interleave is used in the channel coding described above. For this purpose, the digital signal processor 1 feeds the data stream shown in FIG. 8 to a first memory matrix 23 of an interleave at program point 39. The data stream 22 is read in from top left to bottom right in diagonals and the diagonals are offset relative to one another in a downward direction by one storage field. The digital signal processor 1 then reads the data bits line by line from the first memory matrix 23 and sends main to the transmitter unit 2.
Instead of dividing the source-coded image data of the macroblock into a plurality of classes as was described with reference to FIG. 9, it may be advantageous to divide the source-coded image data of an individual macroblock or several macroblocks into a class and to provide every class with a first information word. This reduces the coding delay, although an accurately graduated division into classes is not effected and in certain cases the data rate is accordingly not utilized as efficiently.
FIG. 10 shows a division of an image into regions of different classes designated by numbers. Those classes with a small number correspond to a high degree of error protection. The macroblocks which are arranged in the peripheral regions are transmitted with a lower degree of error protection since these regions of the image play a subordinate part in the subjective perception of the image quality. However, the central regions of the image are associated with the classes having a high degree of error protection, since poor image quality is immediately apparent in these regions.
The process shown in FIG. 9 is especially simplified in that the source-coded image data of an image are divided into classes and these classes are provided with a first information word in each instance. In this way, the image data are combined in larger classes, which results in a particularly simple but less accurate process.
The program flow for receiving, and decoding channel-coded and source-coded image data is shown schematically in FIG. 11. At program point 80, the image data are received by a receiver unit 4, frequency-demodulated and fed to the second digital signal processor 3. The second digital signal processor 3 subjects the digital image data to a convolutional deinterleaving process at program point 81 as is shown in FIG. 8 Interleaving and deinterleaving processes are known and described, among others, eg., in Clark, "Error Correction Coding for Digital Communications", Plenum Press, New York, 1981, ISBN 0-306-40615-2, 345 ff.
The digital image data are read into a first memory matrix 24 in diagonals from bottom left to top right. The second digital signal processor 3 knows that the bits of the synchronizing word are to be found in a determined column of the second memory matrix 24 at fixed intervals, i.e., in a spread. The time spread of the bits of the synchronizing word reduces the susceptibility of the synchronization to burst errors in the data channel. The second digital signal processor 3 determines the synchronizing word via a cross correlation comparison with an ideal synchronizing bit sequence and decides on the basis of an error threshold whether or not the synchronizing word was detected. Cross correlation procedures are known and described, e.g., in Proakis, "Digital Communications", McGraw-Hill, New York, 1989, 141 ff
The time delay occurring in the deinterleaver is used by the second digital signal processor 3 to carry out synchronization. The first bit of the synchronizing word indicates the start of a data frame for the data of all entire image. The image data are then read out of the second matrix memory 24 line by line starting with the first bit of the synchronizing word. The second digital processor 3 knows that a first or second information word 9, 7 follows the first bit of the synchronizing word and the first or second information word 9, 7 has a predetermined number of data bits and is coded within a predetermined code rate. The second digital signal processor 3 carries Out a channel decoding based on the Viterbi algorithm (maximum likelihood sequence estimation) of the first or second information word 9, 7 and determines from the data stored in the first or second information word the quantity of data bits contained in the following code words and image data and the code rates at which the code words or image data bits are to be channel-decoded.
The maximum likelihood sequence estimation process is known and described in Clark, "Error Correction Coding for Digital Communications", Plenum Press, New York, 1981, ISBN 0-306-40615-2, 231 ff., and in Proakis, "Digital Communications", McGraw-Hill, New York, 1989, 394 ff.
At program point 82, the second digital signal processor 3 carries out a channel decoding of the code words 8 based on the Viterbi algorithm. The data stored in the code words 8 are stored by the second digital signal processor 3. The first information word 9 is then channel-decoded corresponding to the fixed code rate by the second digital signal processor 3 based on the Viterbi algorithm. From the data stored in the first information word 9, the second digital signal processor 3 learns how many data bits are contained in the following image data and the code rate at which the following bit data were channel-coded. The second digital signal processor 3 then carries out a channel decoding at program point 83, based on the Viterbi algorithm, of the following, image data bits using, the code rate stored in the first information word 9. Another first information word and additional image data bits are then channel-decoded by the second digital signal processor 3 based on the Viterbi algorithm until all of the image data bits belonging, to an image have been channel-decoded.
At program point 84, the image data bits of an image are subjected to an image source decoding by the second digital signal processor 3 and the channel-decoded and source-decoded image data are then sent to a second data bus 29 by the second digital signal processor 3. Corresponding image source decoding processes are given in an analogous manner by the source-coded image data.
During the image source decoding, the information stored in the code words 8 is used by the second digital signal processor 3. In particularly simple processes, no code words 8 are used so that the second signal processor 3 determines the source-decoded image data, and accordingly the image, from the first information words 9 and the source-coded image data 10.
The source decoded image data are then fed via the second data bus 29 to an image reproduction unit which presents the image at program point 85, whereupon the data of another image is processed by the second digital signal processor 3.
Further, it is not necessary to carry out a deinterleaving process on the reception side in technically simple processes with low image quality in which no interleaving process has been used on the transmission side.
The described processes are used for the transmission of images in mobile telephone communications, for instance. The European digital cordless telephone network (DECT), for example, serves as a transmission channel.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4188616 *||May 15, 1978||Feb 12, 1980||Sony Corporation||Method and system for transmitting and receiving blocks of encoded data words to minimize error distortion in the recovery of said data words|
|US4275457 *||Oct 22, 1979||Jun 23, 1981||Martin Marietta Corporation||Apparatus and method for receiving digital data at a first rate and outputting the data at a different rate|
|US4635262 *||Jun 11, 1984||Jan 6, 1987||U.S. Philips Corporation||Method of detecting synchronization errors in a data transmission system using a linear block code|
|US4692816 *||Jul 2, 1984||Sep 8, 1987||Victor Company Of Japan Ltd.||Digital data transmitting system for transmitting digital data a number of times depending on an information content of the digital data|
|US5321521 *||Jul 16, 1992||Jun 14, 1994||Ricoh Company, Ltd.||Image data compression transmission method utilizing a synchronization code|
|US5335016 *||Jan 23, 1992||Aug 2, 1994||Olympus Optical Co., Ltd.||Image data compressing/coding apparatus|
|US5502571 *||Nov 4, 1993||Mar 26, 1996||U.S. Philips Corporation||Device for processing digital signals first coded by means of variable length coding and device for inversely processing signals thus processed|
|US5602875 *||Jan 13, 1995||Feb 11, 1997||Motorola, Inc.||Method and apparatus for encoding and decoding information in a digtial communication system|
|US5603096 *||Jul 11, 1994||Feb 11, 1997||Qualcomm Incorporated||Reverse link, closed loop power control in a code division multiple access system|
|US5623511 *||Nov 28, 1994||Apr 22, 1997||Lucent Technologies Inc.||Spread spectrum code pulse position modulated receiver having delay spread compensation|
|1||Bosch Technische Berichte, No. 54, 1991, "Uebertragung von Bewegt-und Standbildern . . . ", Dietmar BIERE (No Translation).|
|2||*||Bosch Technische Berichte, No. 54, 1991, Uebertragung von Bewegt und Standbildern . . . , Dietmar BIERE (No Translation).|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US6198849 *||Apr 20, 1996||Mar 6, 2001||Robert Bosch Gmbh||Process for producing and evaluating a video data stream for video transmission|
|US6405338 *||Feb 11, 1998||Jun 11, 2002||Lucent Technologies Inc.||Unequal error protection for perceptual audio coders|
|US6636748 *||Jan 12, 1998||Oct 21, 2003||David A. Monroe||Method and apparatus for image capture, compression and transmission of a visual image over telephone or radio transmission system|
|US6705777||Aug 23, 2002||Mar 16, 2004||Eastman Kodak Company||System and method for digital film development using visible light|
|US6793417||Jan 21, 2003||Sep 21, 2004||Eastman Kodak Company||System and method for digital film development using visible light|
|US6816184 *||Apr 15, 1999||Nov 9, 2004||Texas Instruments Incorporated||Method and apparatus for mapping a location from a video image to a map|
|US7039107||Sep 1, 1999||May 2, 2006||Siemens Aktiengesellschaft||Method and array for processing a digitized image with image points|
|US7583693 *||Jul 17, 2001||Sep 1, 2009||Koninklijke Philips Electronics N.V.||Signal coding|
|US7603610 *||Jan 20, 2003||Oct 13, 2009||Koninklijke Philips Electronics N.V.||Coding a video data stream with unequal error protection based activity|
|US8150124 *||Jan 8, 2010||Apr 3, 2012||Capso Vision Inc.||System and method for multiple viewing-window display of capsule images|
|US20020012360 *||Jul 17, 2001||Jan 31, 2002||Stefano Olivieri||Signal coding|
|US20020034225 *||Jul 17, 2001||Mar 21, 2002||Martini Maria Giuseppina||Coding of a data stream|
|US20030142975 *||Jan 21, 2003||Jul 31, 2003||Edgar Albert D.||System and method for digital film development using visible light|
|US20040080608 *||Dec 20, 2002||Apr 29, 2004||Monroe David A.||Method and apparatus for image capture, compression and transmission of a visual image over telephonic or radio transmission system|
|US20060090118 *||Jan 20, 2003||Apr 27, 2006||Stefano Olivieri||Coding a data stream with unequal error protection|
|US20070076242 *||Sep 26, 2006||Apr 5, 2007||Yasunori Kanai||Printing apparatus, job processing method, printing system, storage medium, and program|
|US20110085021 *||Oct 12, 2009||Apr 14, 2011||Capso Vision Inc.||System and method for display of panoramic capsule images|
|US20110085022 *||Jan 8, 2010||Apr 14, 2011||Capso Vision, Inc.||System and method for multiple viewing-window display of capsule images|
|WO2000019373A1 *||Sep 1, 1999||Apr 6, 2000||Siemens Aktiengesellschaft||Method and array for processing a digitized image with image points|
|WO2001052563A2 *||Jan 12, 2001||Jul 19, 2001||Simoco International Limited||Radio communications system|
|WO2001052563A3 *||Jan 12, 2001||Dec 20, 2001||Alister John Bailey||Radio communications system|
|U.S. Classification||358/426.02, 375/E07.279, 375/E07.222, 375/E07.28, 341/52, 375/E07.267, 358/426.03, 375/E07.211|
|International Classification||H04N19/00, H04N19/89, H04N19/61, H04N7/52, H04N7/26, H04L1/00, H04N7/66, H04N7/64, H04N7/50|
|Cooperative Classification||H04N19/70, H04N19/00, H04N19/89, H04N19/61, H04N7/52, H04L2001/0098|
|European Classification||H04N7/64, H04N7/52, H04N7/50, H04N7/50M, H04N7/66|
|Dec 12, 2001||FPAY||Fee payment|
Year of fee payment: 4
|Dec 14, 2005||FPAY||Fee payment|
Year of fee payment: 8
|Jan 7, 2008||AS||Assignment|
Owner name: IPCOM GMBH & CO. KG, GERMANY
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:ROBERT BOSCH GMBH;REEL/FRAME:020325/0053
Effective date: 20071126
Owner name: IPCOM GMBH & CO. KG,GERMANY
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:ROBERT BOSCH GMBH;REEL/FRAME:020325/0053
Effective date: 20071126
|Dec 16, 2009||FPAY||Fee payment|
Year of fee payment: 12
|May 16, 2013||AS||Assignment|
Owner name: KAROLS DEVELOPMENT CO LLC, NEW YORK
Free format text: SECURITY AGREEMENT;ASSIGNOR:IPCOM GMBH & CO. KG;REEL/FRAME:030427/0352
Effective date: 20080403
|Jun 7, 2013||AS||Assignment|
Owner name: LANDESBANK BADEN-WUERTTEMBERG, GERMANY
Free format text: SECURITY AGREEMENT;ASSIGNOR:IPCOM GMBH & CO. KG;REEL/FRAME:030571/0649
Effective date: 20130607
|Jun 17, 2013||AS||Assignment|
Owner name: ROBERT BOSCH GMBH, GERMANY
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:VOGT, WILHELM;LAPPE, DIRK;PELZ, RODOLFO MANN;AND OTHERS;REEL/FRAME:030626/0933
Effective date: 19951107